解决U形装配线平衡调度问题的免疫协同进化算法  

An Immune Co-evolution Algorithm for Balancing and Sequencing on Mixed-Model U-Lines

作　　者：刘冉[1] 楼佩煌[1] 唐敦兵[1] 杨雷[2] 

机构地区：[1]南京航空航天大学,南京210016 [2]江苏天奇物流系统工程股份有限公司,无锡214187

出　　处：《中国机械工程》2010年第7期815-821,共7页China Mechanical Engineering

基　　金：霍英东教育基金会青年教师基金资助项目(111056);江苏省重大科技成果转化专项资金资助项目(BA2007034);江苏省高校科技成果产业化推进项目(JH07-005);教育部新世纪优秀人才支持计划资助项目(NCET-08)

摘　　要：研究了混流U形装配线平衡与调度的多目标集成优化问题,提出了一种基于Pareto最优的多目标克隆免疫协同进化算法。该算法以两个单克隆抗体群对应平衡与调度两个子问题,分别编码并协同进化,以一个多克隆抗体群保存最优完整解并采取精英策略,使得子种群间既存在协作也存在竞争。提出从抗体的基因型和表现型同时评价抗体亲和度,并改进了共生伙伴选择机制以提高算法的收敛性能。仿真实例证明算法有着更快的收敛速度且比单种群进化算法更适于U形装配线平衡调度问题的求解。The multi--objective optimization problem of balancing and scheduling on mixed-model U-lines had been studied. An immune co--evolutionary algorithm based on Pareto front had been proposed. Two monoclonal antibody populations, which were coded differently and co--evolve with each other,had been designed according to the sub problems of balancing and scheduling. A polyclonal antibody population was used to save the optimal complete solutions and the elitism was executed so that the sub populations compete as they co--evolve. The antibody affinity was evaluated from the phenotype as well as the genotype and the collaboration formation mechanism had been ameliorated to enhance the performance of the algorithm. The algorithm has been proved to have a better performance of convergence and be more suitable for the proposed problem by comparison of the results of three series of experiments from different aspects.

关 键 词：混流U形装配线 平衡与调度 多目标协同进化 免疫算法 

分 类 号：TH16[机械工程—机械制造及自动化] TP39[自动化与计算机技术—计算机应用技术]